Asked — Edited

Pandorabot Not Understanding

Hi guys,

I have a JD (John Dorian) and love everything so far but I cannot get him to be as intelligent as I'd like because I'm a bit stuck. I'll explain the best I can:

Firstly, I didn't mark this as needing assistance as I assume it will be extremely easy for most of you pros to answer. Secondly, I just want to make it clear that I have spent a lot of time training my laptop with speech recognition and although it doesn't have the best microphone, JD generally picks up what I say with confidence levels usually around 80-90.

I have loaded my own pandorabot into using the control for it and adding in my own bots Id but he literally understands nothing. Not even "hello" although I have added this to be the only thing my bot currently knows. I also tried adding another bots Id created by someone else and although it works well when typing my input, usuing my voice he understands nothing.

I have searched relentlessly for a solution as I know it is frowned upon to post regarding an easy fix but I've had no luck.

Any help of advice would be massively appreciated and any more info need will be gladly provided.

Thanks in advance,



Upgrade to ARC Pro

Stay on the cutting edge of robotics with ARC Pro, guaranteeing that your robot is always ahead of the game.


Hi! The pandora bot control uses the speech recognition engine. It's important to train the computer to understand your voice in order to get higher resolution dictation. You can find out more here:

Oh, and also it is important to use a high quality microphone headset:D


Hi DJ,

Yeah, I figured as much!

"This control uses the entire local language dictionary, so a headset or nice microphone is necessary."

I imagine it's my rubbish microphone so I'll focus on some other stuff and come back to this when I get paid! I'll start with getting a better mic and then even more voice training.

Thanks for the super speedy reply,

Much appreciated:)


Nah, its the windows voice recognition that is the problem. Ive used everything from a cheap 10 dollar mic to a 150 dollar stage mic and everything in between with terrible results. Ive spent alot of time training it with my voice and had very marginal improvements. I eventually gave up but much later i tried, just to test the aiml files with googles service through myrobotlab and it worked nearly perfect even through my laptops internal mic which is garbage really. Its too bad because its such a cool feature to have with your bot. Its not the fault of ez robot. Microsoft is all powerful yet a terrible mess as usual. Window 10 in particular is a sad state as far as im concerned. I only use windows at all so as to run my ARC software. All that said i love my ez-b and still use it daily. If im wrong about this i would love someone to explain how they made it work with acceptable results.


windows speech recognition works great with a proper mic, such as a lav mic. The google service is too costly to use

User-inserted image

United Kingdom

To chime in (a rarity these days I know, but I still lurk!)

Windows SAPI is perfectly acceptable if used correctly and trained. I use Windows SAPI for Jarvis and have zero problems.

Without training it wasn't great, after (a lot of) training it has no problems. It has few problems in dictation mode now (although that's about 5 years of constant training).

Google's SAPI isn't exactly great. Neither my phone nor my smart TV's voice controls are anywhere near as accurate as Jarvis is now.


Ok i can understand the cost issues but as far as windows speech recognition working perfectly i cant see how this is possible. I should maybe clarify further that i am referring to pandorabots in particular. I can store voice commands in ARC and it works just fine but when you get into thousands of words..essentially the entire english language then it just falls apart. Gets it right maybe 1 in 10 times. I think its the subtle difference between so many words that it cant cope with. Also my friend has jarvis on his system and it works wonderfully but he cannot get acceptable results from pandorabots in ARC either. As i mentioned in my first post, I did indeed spend many hours with voice training(certainly not 5 years worth but daily for several weeks with marginal improvement. I tried about 6 different mics varying in quality with little improvement as well. The ends simply dont meet here if im getting good results with internal mike in google. Guys im honestly just looking for help here,not complaining :) has anyone manged to have conversations with their pandorabot in ARC? I cant find a single working example anywhere. If someone has a video i would love to see it, if just to give me a glimmer of hope lol if not then ok, no problem, i can live without the feature but i dont want to waste too much time on futile endeavor..any takers? Challenge accepted? Lol


I should add that ive been mucking about with this on and off for about a year with no success and i spent 5 minutes setting it up with googles service. Thats where the confusion lies.


I sure didn't type "perfect". I did type "great", because with a proper Lav Mic, it works great:D

If you wish to pay google for the google speech service, then download the plugin and wait for further development.


Sorry for the incorrect wording bad. Thanks for the plugin though, paying google is not something im excited about but chatting with my inmoov is :) i cant bring myself to purchase yet another mic so the google option will consolidate the missing interaction for me.